We must often search cluttered scenes for items of immediate behavioral relevance ? e.g. our food, our car keys, our friend, and so on. In each case, we use a memory of the target object?s features to efficiently guide our search to objects that share some of its features, so that we are not forced to inspect every object in a crowded scene individually. Efficient visual search is critical for efficient visually guided behavior. Although much is known about the biological mechanisms underlying the selection of objects based on spatial location, much less is known about the mechanisms underlying the selection of objects based on features. To design an effective neural prosthesis or to treat people with sensory or attentional impairments, we need a better understanding of feature attention at the systems level. A better understanding of feature attention will also give us more insight into the mechanisms underlying visual working memory and visual memory recall, as these related functions seem to involve at least partially overlapping neural circuits. Until recently, it was unclear if there was any specific brain structure that stored the information about attended features and used it to guide visual processing in the cortex through top-down feedback. We recently obtained evidence for such a site in prefrontal cortex, in a region that we have termed VPA. Our Aims are focused on a better understanding of VPA and the mechanisms by which it interacts with other visual areas during attention to features. In Aim 1, we will use electrical stimulation paired with fMRI to densely map the projections of a wide expanse of lateral prefrontal cortex, including VPA. This prefrontal ?connectome? will show us how VPA relates to other prefrontal circuits, and it will give us the neural wiring diagram for how VPA interacts with other functional regions throughout the brain. The published connectome will also serve as valuable resource for the neuroscience community. Preliminary results from the connectome are already being used to guide our other two Aims. In Aim 2, we will use pharmacological methods to reversibly deactivate VPA, to test our hypotheses that VPA is the source of feedback that modulates processing in area V4 during attention to features such as shape and color. A positive result would be strong evidence in favor of VPAs feedback control of the ventral stream for object recognition. In Aim 3, we will test our hypotheses about the role of VPA in the dorsal stream, during attention to objects based on their direction of motion. Cells in VPA, MT, MST, FST, and LIP will be recorded simultaneously, to test whether neural activity in VPA has the temporal properties needed to support VPA?s causal role in attention to motion. We will then use new technology we have developed to optogenetically suppress VPA and test whether it impairs attention to motion and reduces or eliminates the effects of attention to motion on the responses of cells in areas MT, MST, FST, and LIP. In total, we expect these studies to give us the best account so far of how the interactions of VPA with multiple brain structures leads to effective visual processing during attention to object features.